gf128mul.c 12 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416
  1. /* gf128mul.c - GF(2^128) multiplication functions
  2. *
  3. * Copyright (c) 2003, Dr Brian Gladman, Worcester, UK.
  4. * Copyright (c) 2006, Rik Snel <rsnel@cube.dyndns.org>
  5. *
  6. * Based on Dr Brian Gladman's (GPL'd) work published at
  7. * http://gladman.plushost.co.uk/oldsite/cryptography_technology/index.php
  8. * See the original copyright notice below.
  9. *
  10. * This program is free software; you can redistribute it and/or modify it
  11. * under the terms of the GNU General Public License as published by the Free
  12. * Software Foundation; either version 2 of the License, or (at your option)
  13. * any later version.
  14. */
  15. /*
  16. ---------------------------------------------------------------------------
  17. Copyright (c) 2003, Dr Brian Gladman, Worcester, UK. All rights reserved.
  18. LICENSE TERMS
  19. The free distribution and use of this software in both source and binary
  20. form is allowed (with or without changes) provided that:
  21. 1. distributions of this source code include the above copyright
  22. notice, this list of conditions and the following disclaimer;
  23. 2. distributions in binary form include the above copyright
  24. notice, this list of conditions and the following disclaimer
  25. in the documentation and/or other associated materials;
  26. 3. the copyright holder's name is not used to endorse products
  27. built using this software without specific written permission.
  28. ALTERNATIVELY, provided that this notice is retained in full, this product
  29. may be distributed under the terms of the GNU General Public License (GPL),
  30. in which case the provisions of the GPL apply INSTEAD OF those given above.
  31. DISCLAIMER
  32. This software is provided 'as is' with no explicit or implied warranties
  33. in respect of its properties, including, but not limited to, correctness
  34. and/or fitness for purpose.
  35. ---------------------------------------------------------------------------
  36. Issue 31/01/2006
  37. This file provides fast multiplication in GF(2^128) as required by several
  38. cryptographic authentication modes
  39. */
  40. #include <crypto/gf128mul.h>
  41. #include <linux/kernel.h>
  42. #include <linux/module.h>
  43. #include <linux/slab.h>
  44. #define gf128mul_dat(q) { \
  45. q(0x00), q(0x01), q(0x02), q(0x03), q(0x04), q(0x05), q(0x06), q(0x07),\
  46. q(0x08), q(0x09), q(0x0a), q(0x0b), q(0x0c), q(0x0d), q(0x0e), q(0x0f),\
  47. q(0x10), q(0x11), q(0x12), q(0x13), q(0x14), q(0x15), q(0x16), q(0x17),\
  48. q(0x18), q(0x19), q(0x1a), q(0x1b), q(0x1c), q(0x1d), q(0x1e), q(0x1f),\
  49. q(0x20), q(0x21), q(0x22), q(0x23), q(0x24), q(0x25), q(0x26), q(0x27),\
  50. q(0x28), q(0x29), q(0x2a), q(0x2b), q(0x2c), q(0x2d), q(0x2e), q(0x2f),\
  51. q(0x30), q(0x31), q(0x32), q(0x33), q(0x34), q(0x35), q(0x36), q(0x37),\
  52. q(0x38), q(0x39), q(0x3a), q(0x3b), q(0x3c), q(0x3d), q(0x3e), q(0x3f),\
  53. q(0x40), q(0x41), q(0x42), q(0x43), q(0x44), q(0x45), q(0x46), q(0x47),\
  54. q(0x48), q(0x49), q(0x4a), q(0x4b), q(0x4c), q(0x4d), q(0x4e), q(0x4f),\
  55. q(0x50), q(0x51), q(0x52), q(0x53), q(0x54), q(0x55), q(0x56), q(0x57),\
  56. q(0x58), q(0x59), q(0x5a), q(0x5b), q(0x5c), q(0x5d), q(0x5e), q(0x5f),\
  57. q(0x60), q(0x61), q(0x62), q(0x63), q(0x64), q(0x65), q(0x66), q(0x67),\
  58. q(0x68), q(0x69), q(0x6a), q(0x6b), q(0x6c), q(0x6d), q(0x6e), q(0x6f),\
  59. q(0x70), q(0x71), q(0x72), q(0x73), q(0x74), q(0x75), q(0x76), q(0x77),\
  60. q(0x78), q(0x79), q(0x7a), q(0x7b), q(0x7c), q(0x7d), q(0x7e), q(0x7f),\
  61. q(0x80), q(0x81), q(0x82), q(0x83), q(0x84), q(0x85), q(0x86), q(0x87),\
  62. q(0x88), q(0x89), q(0x8a), q(0x8b), q(0x8c), q(0x8d), q(0x8e), q(0x8f),\
  63. q(0x90), q(0x91), q(0x92), q(0x93), q(0x94), q(0x95), q(0x96), q(0x97),\
  64. q(0x98), q(0x99), q(0x9a), q(0x9b), q(0x9c), q(0x9d), q(0x9e), q(0x9f),\
  65. q(0xa0), q(0xa1), q(0xa2), q(0xa3), q(0xa4), q(0xa5), q(0xa6), q(0xa7),\
  66. q(0xa8), q(0xa9), q(0xaa), q(0xab), q(0xac), q(0xad), q(0xae), q(0xaf),\
  67. q(0xb0), q(0xb1), q(0xb2), q(0xb3), q(0xb4), q(0xb5), q(0xb6), q(0xb7),\
  68. q(0xb8), q(0xb9), q(0xba), q(0xbb), q(0xbc), q(0xbd), q(0xbe), q(0xbf),\
  69. q(0xc0), q(0xc1), q(0xc2), q(0xc3), q(0xc4), q(0xc5), q(0xc6), q(0xc7),\
  70. q(0xc8), q(0xc9), q(0xca), q(0xcb), q(0xcc), q(0xcd), q(0xce), q(0xcf),\
  71. q(0xd0), q(0xd1), q(0xd2), q(0xd3), q(0xd4), q(0xd5), q(0xd6), q(0xd7),\
  72. q(0xd8), q(0xd9), q(0xda), q(0xdb), q(0xdc), q(0xdd), q(0xde), q(0xdf),\
  73. q(0xe0), q(0xe1), q(0xe2), q(0xe3), q(0xe4), q(0xe5), q(0xe6), q(0xe7),\
  74. q(0xe8), q(0xe9), q(0xea), q(0xeb), q(0xec), q(0xed), q(0xee), q(0xef),\
  75. q(0xf0), q(0xf1), q(0xf2), q(0xf3), q(0xf4), q(0xf5), q(0xf6), q(0xf7),\
  76. q(0xf8), q(0xf9), q(0xfa), q(0xfb), q(0xfc), q(0xfd), q(0xfe), q(0xff) \
  77. }
  78. /*
  79. * Given a value i in 0..255 as the byte overflow when a field element
  80. * in GF(2^128) is multiplied by x^8, the following macro returns the
  81. * 16-bit value that must be XOR-ed into the low-degree end of the
  82. * product to reduce it modulo the polynomial x^128 + x^7 + x^2 + x + 1.
  83. *
  84. * There are two versions of the macro, and hence two tables: one for
  85. * the "be" convention where the highest-order bit is the coefficient of
  86. * the highest-degree polynomial term, and one for the "le" convention
  87. * where the highest-order bit is the coefficient of the lowest-degree
  88. * polynomial term. In both cases the values are stored in CPU byte
  89. * endianness such that the coefficients are ordered consistently across
  90. * bytes, i.e. in the "be" table bits 15..0 of the stored value
  91. * correspond to the coefficients of x^15..x^0, and in the "le" table
  92. * bits 15..0 correspond to the coefficients of x^0..x^15.
  93. *
  94. * Therefore, provided that the appropriate byte endianness conversions
  95. * are done by the multiplication functions (and these must be in place
  96. * anyway to support both little endian and big endian CPUs), the "be"
  97. * table can be used for multiplications of both "bbe" and "ble"
  98. * elements, and the "le" table can be used for multiplications of both
  99. * "lle" and "lbe" elements.
  100. */
  101. #define xda_be(i) ( \
  102. (i & 0x80 ? 0x4380 : 0) ^ (i & 0x40 ? 0x21c0 : 0) ^ \
  103. (i & 0x20 ? 0x10e0 : 0) ^ (i & 0x10 ? 0x0870 : 0) ^ \
  104. (i & 0x08 ? 0x0438 : 0) ^ (i & 0x04 ? 0x021c : 0) ^ \
  105. (i & 0x02 ? 0x010e : 0) ^ (i & 0x01 ? 0x0087 : 0) \
  106. )
  107. #define xda_le(i) ( \
  108. (i & 0x80 ? 0xe100 : 0) ^ (i & 0x40 ? 0x7080 : 0) ^ \
  109. (i & 0x20 ? 0x3840 : 0) ^ (i & 0x10 ? 0x1c20 : 0) ^ \
  110. (i & 0x08 ? 0x0e10 : 0) ^ (i & 0x04 ? 0x0708 : 0) ^ \
  111. (i & 0x02 ? 0x0384 : 0) ^ (i & 0x01 ? 0x01c2 : 0) \
  112. )
  113. static const u16 gf128mul_table_le[256] = gf128mul_dat(xda_le);
  114. static const u16 gf128mul_table_be[256] = gf128mul_dat(xda_be);
  115. /*
  116. * The following functions multiply a field element by x^8 in
  117. * the polynomial field representation. They use 64-bit word operations
  118. * to gain speed but compensate for machine endianness and hence work
  119. * correctly on both styles of machine.
  120. */
  121. static void gf128mul_x8_lle(be128 *x)
  122. {
  123. u64 a = be64_to_cpu(x->a);
  124. u64 b = be64_to_cpu(x->b);
  125. u64 _tt = gf128mul_table_le[b & 0xff];
  126. x->b = cpu_to_be64((b >> 8) | (a << 56));
  127. x->a = cpu_to_be64((a >> 8) ^ (_tt << 48));
  128. }
  129. static void gf128mul_x8_bbe(be128 *x)
  130. {
  131. u64 a = be64_to_cpu(x->a);
  132. u64 b = be64_to_cpu(x->b);
  133. u64 _tt = gf128mul_table_be[a >> 56];
  134. x->a = cpu_to_be64((a << 8) | (b >> 56));
  135. x->b = cpu_to_be64((b << 8) ^ _tt);
  136. }
  137. void gf128mul_x8_ble(le128 *r, const le128 *x)
  138. {
  139. u64 a = le64_to_cpu(x->a);
  140. u64 b = le64_to_cpu(x->b);
  141. u64 _tt = gf128mul_table_be[a >> 56];
  142. r->a = cpu_to_le64((a << 8) | (b >> 56));
  143. r->b = cpu_to_le64((b << 8) ^ _tt);
  144. }
  145. EXPORT_SYMBOL(gf128mul_x8_ble);
  146. void gf128mul_lle(be128 *r, const be128 *b)
  147. {
  148. be128 p[8];
  149. int i;
  150. p[0] = *r;
  151. for (i = 0; i < 7; ++i)
  152. gf128mul_x_lle(&p[i + 1], &p[i]);
  153. memset(r, 0, sizeof(*r));
  154. for (i = 0;;) {
  155. u8 ch = ((u8 *)b)[15 - i];
  156. if (ch & 0x80)
  157. be128_xor(r, r, &p[0]);
  158. if (ch & 0x40)
  159. be128_xor(r, r, &p[1]);
  160. if (ch & 0x20)
  161. be128_xor(r, r, &p[2]);
  162. if (ch & 0x10)
  163. be128_xor(r, r, &p[3]);
  164. if (ch & 0x08)
  165. be128_xor(r, r, &p[4]);
  166. if (ch & 0x04)
  167. be128_xor(r, r, &p[5]);
  168. if (ch & 0x02)
  169. be128_xor(r, r, &p[6]);
  170. if (ch & 0x01)
  171. be128_xor(r, r, &p[7]);
  172. if (++i >= 16)
  173. break;
  174. gf128mul_x8_lle(r);
  175. }
  176. }
  177. EXPORT_SYMBOL(gf128mul_lle);
  178. void gf128mul_bbe(be128 *r, const be128 *b)
  179. {
  180. be128 p[8];
  181. int i;
  182. p[0] = *r;
  183. for (i = 0; i < 7; ++i)
  184. gf128mul_x_bbe(&p[i + 1], &p[i]);
  185. memset(r, 0, sizeof(*r));
  186. for (i = 0;;) {
  187. u8 ch = ((u8 *)b)[i];
  188. if (ch & 0x80)
  189. be128_xor(r, r, &p[7]);
  190. if (ch & 0x40)
  191. be128_xor(r, r, &p[6]);
  192. if (ch & 0x20)
  193. be128_xor(r, r, &p[5]);
  194. if (ch & 0x10)
  195. be128_xor(r, r, &p[4]);
  196. if (ch & 0x08)
  197. be128_xor(r, r, &p[3]);
  198. if (ch & 0x04)
  199. be128_xor(r, r, &p[2]);
  200. if (ch & 0x02)
  201. be128_xor(r, r, &p[1]);
  202. if (ch & 0x01)
  203. be128_xor(r, r, &p[0]);
  204. if (++i >= 16)
  205. break;
  206. gf128mul_x8_bbe(r);
  207. }
  208. }
  209. EXPORT_SYMBOL(gf128mul_bbe);
  210. /* This version uses 64k bytes of table space.
  211. A 16 byte buffer has to be multiplied by a 16 byte key
  212. value in GF(2^128). If we consider a GF(2^128) value in
  213. the buffer's lowest byte, we can construct a table of
  214. the 256 16 byte values that result from the 256 values
  215. of this byte. This requires 4096 bytes. But we also
  216. need tables for each of the 16 higher bytes in the
  217. buffer as well, which makes 64 kbytes in total.
  218. */
  219. /* additional explanation
  220. * t[0][BYTE] contains g*BYTE
  221. * t[1][BYTE] contains g*x^8*BYTE
  222. * ..
  223. * t[15][BYTE] contains g*x^120*BYTE */
  224. struct gf128mul_64k *gf128mul_init_64k_bbe(const be128 *g)
  225. {
  226. struct gf128mul_64k *t;
  227. int i, j, k;
  228. t = kzalloc(sizeof(*t), GFP_KERNEL);
  229. if (!t)
  230. goto out;
  231. for (i = 0; i < 16; i++) {
  232. t->t[i] = kzalloc(sizeof(*t->t[i]), GFP_KERNEL);
  233. if (!t->t[i]) {
  234. gf128mul_free_64k(t);
  235. t = NULL;
  236. goto out;
  237. }
  238. }
  239. t->t[0]->t[1] = *g;
  240. for (j = 1; j <= 64; j <<= 1)
  241. gf128mul_x_bbe(&t->t[0]->t[j + j], &t->t[0]->t[j]);
  242. for (i = 0;;) {
  243. for (j = 2; j < 256; j += j)
  244. for (k = 1; k < j; ++k)
  245. be128_xor(&t->t[i]->t[j + k],
  246. &t->t[i]->t[j], &t->t[i]->t[k]);
  247. if (++i >= 16)
  248. break;
  249. for (j = 128; j > 0; j >>= 1) {
  250. t->t[i]->t[j] = t->t[i - 1]->t[j];
  251. gf128mul_x8_bbe(&t->t[i]->t[j]);
  252. }
  253. }
  254. out:
  255. return t;
  256. }
  257. EXPORT_SYMBOL(gf128mul_init_64k_bbe);
  258. void gf128mul_free_64k(struct gf128mul_64k *t)
  259. {
  260. int i;
  261. for (i = 0; i < 16; i++)
  262. kfree_sensitive(t->t[i]);
  263. kfree_sensitive(t);
  264. }
  265. EXPORT_SYMBOL(gf128mul_free_64k);
  266. void gf128mul_64k_bbe(be128 *a, const struct gf128mul_64k *t)
  267. {
  268. u8 *ap = (u8 *)a;
  269. be128 r[1];
  270. int i;
  271. *r = t->t[0]->t[ap[15]];
  272. for (i = 1; i < 16; ++i)
  273. be128_xor(r, r, &t->t[i]->t[ap[15 - i]]);
  274. *a = *r;
  275. }
  276. EXPORT_SYMBOL(gf128mul_64k_bbe);
  277. /* This version uses 4k bytes of table space.
  278. A 16 byte buffer has to be multiplied by a 16 byte key
  279. value in GF(2^128). If we consider a GF(2^128) value in a
  280. single byte, we can construct a table of the 256 16 byte
  281. values that result from the 256 values of this byte.
  282. This requires 4096 bytes. If we take the highest byte in
  283. the buffer and use this table to get the result, we then
  284. have to multiply by x^120 to get the final value. For the
  285. next highest byte the result has to be multiplied by x^112
  286. and so on. But we can do this by accumulating the result
  287. in an accumulator starting with the result for the top
  288. byte. We repeatedly multiply the accumulator value by
  289. x^8 and then add in (i.e. xor) the 16 bytes of the next
  290. lower byte in the buffer, stopping when we reach the
  291. lowest byte. This requires a 4096 byte table.
  292. */
  293. struct gf128mul_4k *gf128mul_init_4k_lle(const be128 *g)
  294. {
  295. struct gf128mul_4k *t;
  296. int j, k;
  297. t = kzalloc(sizeof(*t), GFP_KERNEL);
  298. if (!t)
  299. goto out;
  300. t->t[128] = *g;
  301. for (j = 64; j > 0; j >>= 1)
  302. gf128mul_x_lle(&t->t[j], &t->t[j+j]);
  303. for (j = 2; j < 256; j += j)
  304. for (k = 1; k < j; ++k)
  305. be128_xor(&t->t[j + k], &t->t[j], &t->t[k]);
  306. out:
  307. return t;
  308. }
  309. EXPORT_SYMBOL(gf128mul_init_4k_lle);
  310. struct gf128mul_4k *gf128mul_init_4k_bbe(const be128 *g)
  311. {
  312. struct gf128mul_4k *t;
  313. int j, k;
  314. t = kzalloc(sizeof(*t), GFP_KERNEL);
  315. if (!t)
  316. goto out;
  317. t->t[1] = *g;
  318. for (j = 1; j <= 64; j <<= 1)
  319. gf128mul_x_bbe(&t->t[j + j], &t->t[j]);
  320. for (j = 2; j < 256; j += j)
  321. for (k = 1; k < j; ++k)
  322. be128_xor(&t->t[j + k], &t->t[j], &t->t[k]);
  323. out:
  324. return t;
  325. }
  326. EXPORT_SYMBOL(gf128mul_init_4k_bbe);
  327. void gf128mul_4k_lle(be128 *a, const struct gf128mul_4k *t)
  328. {
  329. u8 *ap = (u8 *)a;
  330. be128 r[1];
  331. int i = 15;
  332. *r = t->t[ap[15]];
  333. while (i--) {
  334. gf128mul_x8_lle(r);
  335. be128_xor(r, r, &t->t[ap[i]]);
  336. }
  337. *a = *r;
  338. }
  339. EXPORT_SYMBOL(gf128mul_4k_lle);
  340. void gf128mul_4k_bbe(be128 *a, const struct gf128mul_4k *t)
  341. {
  342. u8 *ap = (u8 *)a;
  343. be128 r[1];
  344. int i = 0;
  345. *r = t->t[ap[0]];
  346. while (++i < 16) {
  347. gf128mul_x8_bbe(r);
  348. be128_xor(r, r, &t->t[ap[i]]);
  349. }
  350. *a = *r;
  351. }
  352. EXPORT_SYMBOL(gf128mul_4k_bbe);
  353. MODULE_LICENSE("GPL");
  354. MODULE_DESCRIPTION("Functions for multiplying elements of GF(2^128)");